[borgel]

Wow, an ASIC is a big step! I'm impressed you needed to hit that level of (cost down, I assume?). Or was it to get better HW security, perhaps to use it for billing?

Anyway, I had an original Bluetooth Automatic until I upgraded to a Pro when it was launched. I love them! Sorry to hear this is the end.

[mrlambchop]

Vehicle compatibility was really the key - building a cross bar that allowed any pin to be mapped for any function (i.e. CAN on any combo of pins). We included a hardware protocol decoder and a ton of electrical safety items as well as security and reliability tools. 200MHz processor, 2MB flash and 256kB of ram in a single QFN package. A great project to work on! One career highlight for sure.

[borgel]

Ahh, interesting. I can see how that combination of features would be easier to fit in an ASIC than as discreet components in the tiny dongle. Pretty cool hardware combination!

Can you comment on what the safety/security/reliability blocks were?

[mrlambchop]

A couple of examples of hard problems might be the best approach.

- loss of ground

A vehicle OBD connector can at times, loose its GND pin. There are actually two GND pins - chassis and electrical that can sometimes disagree. Its possible for one of either of them to either (a) loose connectivity (b) bias towards the signal level or head towards a negative signal level (-2v was our typical observation).

To prevent an OBD protocol pin from dumping current down the channel, a complex HW/SW solution is needed to monitor and react quickly to prevent downstream ECUs from being fried.

- security

We built a custom debug tool based on CM-DAP (ARM debugger technology) that included OOB security paths to unlock a device for re-flashing / debugging. I think we made 500-1k of these in total for manufacturing/RMA/logistic purposes. This was super fun to work on :)